To type or not to type: the use of unaffected siblings in nonparametric linkage analysis

Unaffected individuals are often disregarded in nonparametric linkage analysis. Because of the presumed high complexity of genetic interactions and the resulting low penetrance of any single genetic effect, the statistical contribution of unaffected sib pairs is thought to be considerably lower than that of the affected. However, in cases where a large number of unaffected family members are available for genotyping--as is the case in the simulated Genetic Analysis Workshop 12 data set--the relatively low value of the unaffected sibs may be offset by their number. In this paper, I demonstrate that unaffected siblings do contain a significant amount of statistical information, and show that the most powerful method of nonparametric analysis utilizes both affected and unaffected individuals. I also show that selecting the unaffected individuals based on their age is the most cost effective and usually the most powerful strategy.